> but that is literally how scientific fields always approach arbitrary classifications

No, not always. What you are describing there are shoddy classifications. I already addressed this exact point in my previous reply when I stated that classifications are crafted to serve a useful purpose. It is sometimes useful to hew close to the historical norm; other times it serves no purpose.

> But remember, it must not take into account what we today consider "planets" to be.

An arbitrary restriction that I do not accept.

> please give me an example of what kind of definition for "planet" you'd be happy with

I already did this in my original comment. I proposed an earth centric system that bins mass by powers of ten. My proposed range for planets was 10^23 <= x < 10^26. Hydrostatic equilibrium works well as a constraint to eliminate things that are clearly not earth like (such as black holes) but alternative formulations such as stipulating an acceptable density range are possible (such cross cutting subcategories might even be preferable).

Note that this would result in ganymede being classified as both a planet and a moon. I think that makes perfect sense given how similar in size it is to mars. In fact ganymede being excluded as a planet is precisely the sort of gerrymandering I object to.

As it turns out my proposed criteria also relegates pluto to the status of dwarf planet. I think that's sensible given that it's both smaller and lighter than the moon. However I'd be willing to entertain a set of criteria that acknowledged both pluto and the moon as proper planets - the important thing to me is consistency.

I don't see how your proposed criteria are any less of a shoddy retcon.